Environment perception method for autonomous navigation of distributed heterogeneous multi-robot system

Abstract

The application discloses an environment perception method for autonomous navigation of a distributed heterogeneous multi-robot system. Firstly, each robot initializes to obtain its own 2D radar information, position information and real-time safety radius, and broadcasts its real-time position information and safety radius in real time. Then, the position information and safety radius of other robots are obtained through communication between the robots. Finally, the position information of the robot, the position information of other robots and the safety radius are converted into pseudo-radar data, and then are fused with the 2D radar information of the robot itself, so that the state representation of environment perception is completed. The fused environment perception state representation is taken as the state space of the robot, is input into a neural network of a control strategy, the neural network model is trained through reinforcement learning, and the robot action is output, and the above steps are repeated until the robot reaches the target point. Through the above steps, the environment perception problem that the heterogeneous robot cannot correctly represent the actual safety radius of other robots is solved.

Description

Technical Field

[0001] This invention relates to the field of environmental perception for autonomous robot navigation, and in particular to an environmental perception method for autonomous navigation of distributed heterogeneous multi-robot systems. Background Technology

[0002] Autonomous robot navigation refers to the ability of a robot to autonomously perceive, locate, plan paths, and reach a target position within its environment. Environmental perception is the foundation of autonomous robot navigation. Robots need to use various sensors to acquire information about their environment, including maps, obstacles, boundaries, and the positions of other objects. 2D radar is a commonly used sensor for autonomous robot navigation and environmental perception. It has high-precision distance measurement capabilities, accurately measuring the distance between the robot and surrounding objects. Furthermore, 2D radar exhibits strong anti-interference capabilities, providing stable and reliable perception data even under various complex environmental conditions. Most importantly, the relatively low cost of 2D radar makes it an economical and effective choice, widely used in the field of robot navigation.

[0003] In industry, robots differ in shape and structure due to their varying functions, resulting in different safe radii during operation. The safe radii also vary depending on the goods being transported. When a robot is equipped with a lightweight 2D radar sensor, the radar beam can only scan a single plane and cannot accurately detect the true safe radii of other robots, leading to erroneous perception and potential collisions. Therefore, an effective method is needed to convert the robot's relative position information into pseudo-radar data. Summary of the Invention

[0004] To address this issue, this invention proposes an environmental perception method for autonomous navigation in distributed heterogeneous multi-robot systems. This method addresses the limitation of 2D radar ray scanning only on a single plane and failing to accurately detect the true safety radius of other robots. The robot transforms its position information into pseudo-radar data, which is then fused with the robot's own radar data as navigation state input, thus converting different types of perception information into a unified type.

[0005] An environmental perception method for autonomous navigation of distributed heterogeneous multi-robot systems, characterized by the following steps:

[0006] S1: Initialize robot H, obtain its own 2D radar information, position information and real-time safety radius, and then broadcast its own position information and real-time safety radius;

[0007] S2: Calculate the pseudo radar data of other robots relative to robot H using the position information of robot H, the position information of other robots, and the safety radius of other robots;

[0008] S3: Integrate the pseudo radar data of other robots relative to robot H and the radar information of robot H itself to obtain the state representation of robot H's environmental perception, and use the state representation of robot H as the state space of robot H.

[0009] S4: Input the state space into the neural network of the control strategy, train the neural network model through reinforcement learning, and output robot H action a. Robot H executes action a.

[0010] S5: Determine whether robot H has reached the target point. If it has reached the target point, navigation ends. If it has not reached the target point, continue to step S1.

[0011] Furthermore, in step S1, the 2D radar information of robot H includes the distance M = {m1, m2, m3, ...} and the azimuth angle. The position information of robot H includes its coordinates (x, y) in the global coordinate system and the angle θ between the robot H's forward direction and the positive direction of the global coordinate system. H .

[0012] Furthermore, the feature is that step S2, calculating the pseudo radar data of other robots relative to robot H using the position information of robot H, the position information of other robots, and the safety radius of other robots, includes the following steps:

[0013] S21: Calculate the distance and angle between robot H and other robots based on their position information and the position information of other robots. The position information of other robots is (x i ,y i ), i = 1, 2, 3…, with a safety radius of r i The calculation method is as follows:

[0014]

[0015] Where θ i Di is the angle between the line connecting robot H and other robots and the positive direction of the global coordinate system, and Di is the distance between robot H and other robots.

[0016] S22: Establish a rectangular coordinate system with robot H as the center and the forward direction of robot H as the positive x-axis. Therefore, the angle between the line connecting the other robots in the robot H coordinate system and the positive x-axis is:

[0017]

[0018] The equation of the circle formed by the other robot safety radii in the robot's H coordinate system is:

[0019]

[0020] Transform the equation of the curve into polar coordinates, and let:

[0021]

[0022] Therefore, the curve equations for other robots in the polar scale system are:

[0023]

[0024] Solving the equation yields:

[0025]

[0026] S23: The outlines of other robots detected by robot H's 2D radar scan may differ from the arcs formed by the actual safety radii. Using the arcs formed by the safety radii of other robots as their outlines can more accurately represent robot H's perception information. The equation of the arc curves formed by the safety radii of other robots is:

[0027]

[0028] The range of values ​​for the equation of the circular arc curve is: The calculation formula is:

[0029]

[0030]

[0031] Starting from robot H, draw a tangent line that forms a circle with the safety radii of the other robots. Let H be the angle formed by the line connecting robot H and the other robots and this tangent;

[0032] S24: Assuming robot H's radar scan detects the arc formed by the safety radii of other robots, based on the azimuth angle of robot H's 2D radar information... Find the pseudo-radar data for the azimuth angle of robot H, if the distance between other robots and robot H is less than R. lidar ,but

[0033]

[0034] in, Represents the azimuth angle of the i-th robot relative to robot H. Pseudo-radar data at the location, R lidarThe distance between robot H and robot H represents the maximum radar detection radius. If the distance between the i-th robot and robot H is greater than the maximum radar detection radius R of robot H, then... lidar Then the pseudo radar data for the azimuth angle corresponding to robot H are all R. lidar Therefore, the pseudo-radar data for the azimuth angle of the i-th robot relative to robot H is finally:

[0035]

[0036] in Represents the azimuth angle of the i-th robot relative to robot H. The pseudo-radar data at point i, therefore the pseudo-radar data of the i-th robot relative to robot H can be represented as:

[0037] S25: Fusion of pseudo-radar data of all other robots relative to robot H, i.e., finding the minimum value of the upper radar data of other robots relative to robot H at the corresponding azimuth angle:

[0038]

[0039] in This represents the azimuth angle of the i-th robot at robot H. Pseudo-radar data at the location, Represents the azimuth angle of all other robots relative to robot H. The minimum value of the pseudo-radar data at that location, therefore the fused pseudo-radar data of other robots relative to robot H is:

[0040] Preferably, in step S3 above, pseudo radar data of other robots relative to robot H and radar information of robot H itself are fused:

[0041]

[0042] Where m j This represents the robot H itself in the azimuth angle. 2D radar range information at L fusion This represents the state representation of the robot's environmental perception after fusing pseudo-radar data from H with its own 2D radar distance information.

[0043] The beneficial effects of this invention compared to the prior art are as follows:

[0044] 2D radar rays can only scan objects on a single plane and cannot accurately scan the true safe radius of other robots. This invention solves this problem by converting robot position information and safe radius into pseudo radar data.

[0045] This invention achieves the homogenization of position information data and radar data by converting robot position information and safety radius into pseudo radar data, and reduces the dimensionality of neural network state input by fusion, thereby reducing the difficulty of neural network training. Attached Figure Description

[0046] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0047] Figure 1 This is a flowchart of the environmental perception method for autonomous navigation of distributed heterogeneous multi-robot systems according to the present invention.

[0048] Figure 2 A schematic diagram of the mathematical model for converting robot position information and safety radius into pseudo-radar data provided by this invention.

[0049] Figure 3 This is a schematic diagram illustrating the process of converting robot position information and safety radius into pseudo radar data, as provided by the present invention. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0051] Please see Figure 1 An embodiment of the present invention provides an environmental perception method for autonomous navigation of a distributed heterogeneous multi-robot system, comprising the following steps:

[0052] S1: Initialize robot H, acquire its own 2D radar information, position information, and real-time safe radius, and then broadcast its own position information and real-time safe radius. Robot H's 2D radar information includes distance M = {m1, m2, m3, ...} and azimuth angle. The position information of robot H includes its coordinates (x, y) in the global coordinate system and the angle θ between the robot H's forward direction and the positive direction of the global coordinate system. H The location information of other robots is (x i ,y i ), i = 1, 2, 3…, with a safety radius of ri .

[0053] S2: Calculating the pseudo radar data of other robots relative to robot H using robot H's position information, other robot position information, and other robots' safety radii includes the following sub-steps:

[0054] S21: In the global coordinate system, calculate the distance and angle between robot H and other robots based on their position information. The calculation method is as follows:

[0055]

[0056] Where θ i Di is the angle between the line connecting robot H and other robots and the positive direction of the global coordinate system, and Di is the distance between robot H and other robots.

[0057] S22: Establish a Cartesian coordinate system with robot H as the center and the direction of robot H's movement as the positive x-axis. Please refer to [link / reference]. Figure 2 Therefore, the angle between the line connecting the other robots in the robot's H coordinate system and the positive x-axis is:

[0058]

[0059] The equation of the circle formed by the other robot safety radii in the robot's H coordinate system is:

[0060]

[0061] Transform the equation of the curve into polar coordinates, and let:

[0062]

[0063] Therefore, the curve equations for other robots in the polar scale system are:

[0064]

[0065] Solving the equation yields:

[0066]

[0067] S23: The outlines of other robots detected by robot H's 2D radar scan may differ from the arcs formed by the actual safety radii. Using the arcs formed by the safety radii of other robots as their outlines can more accurately represent robot H's perception information. The equation of the arc curves formed by the safety radii of other robots is:

[0068]

[0069] The range of values ​​for the equation of the circular arc curve is: The calculation formula is:

[0070]

[0071]

[0072] Starting from robot H, draw a tangent line that forms a circle with the safety radii of the other robots. Let H be the angle formed by the line connecting robot H and the other robots and this tangent;

[0073] S24: Assuming robot H's radar scan detects an arc formed by the safety radius of other robots, please refer to... Figure 3 Based on the azimuth angle of robot H's 2D radar information Find the pseudo-radar data for the azimuth angle of robot H, if the distance between other robots and robot H is less than R. lidar ,

[0074] but

[0075]

[0076] in, Represents the azimuth angle of the i-th robot relative to robot H. Pseudo-radar data at the location, R lidar The distance between robot H and robot H represents the maximum radar detection radius. If the distance between the i-th robot and robot H is greater than the maximum radar detection radius R of robot H, then... lidar Then the pseudo radar data for the azimuth angle corresponding to robot H are all R. lidar Therefore, the pseudo-radar data for the azimuth angle of the i-th robot relative to robot H is finally:

[0077]

[0078] in Represents the azimuth angle of the i-th robot relative to robot H. The pseudo-radar data at point i, therefore the pseudo-radar data of the i-th robot relative to robot H can be represented as:

[0079] S25: Fusion of pseudo-radar data of all other robots relative to robot H, i.e., finding the minimum value of the upper radar data of other robots relative to robot H at the corresponding azimuth angle:

[0080]

[0081] in This represents the azimuth angle of the i-th robot at robot H. Pseudo-radar data at the location, Represents the azimuth angle of all other robots relative to robot H. The minimum value of the pseudo-radar data at that location, therefore the fused pseudo-radar data of other robots relative to robot H is:

[0082] S3: Integrate the pseudo radar data of other robots relative to robot H and the radar information of robot H itself to obtain the state representation of robot H's environmental perception, and use the state representation of robot H as the state space of robot H.

[0083] In step S3 above, pseudo radar data from other robots relative to robot H and radar information from robot H itself are fused:

[0084]

[0085] Where m j This represents the robot H itself in the azimuth angle. 2D radar range information at L fusion This represents the state representation of the robot's environmental perception after fusing pseudo-radar data from H with its own 2D radar distance information.

[0086] S4: Input the state space into the neural network of the control strategy, train the neural network model through reinforcement learning, and output robot H action a. Robot H executes action a.

[0087] S5: Determine whether robot H has reached the target point. If it has reached the target point, navigation ends. If it has not reached the target point, continue to step S1.

[0088] In summary, the environmental perception method for autonomous navigation of distributed heterogeneous multi-robot systems provided by this invention involves the following steps: First, each robot initializes by acquiring its own 2D radar information, position information, and real-time safe radius, and continuously broadcasts its real-time position information and real-time safe radius. Then, it receives the position information and safe radius of other robots through inter-robot communication. Finally, it converts its own position information, the position information of other robots, and the safe radii of other robots into pseudo-radar data, and then fuses this data with the robot's own 2D radar information to obtain fused pseudo-radar data. The fused pseudo-radar data is used as the robot's state space, which is then input into a neural network to map robot actions. After executing an action, the above steps are repeated until the robot reaches the target point. Heterogeneous robots have different shapes and safe radii, and the safe radius changes continuously as the load changes during task execution. When a robot is equipped with 2D radar, the radar beam can only scan a plane of objects and cannot correctly scan the true safe radii of other robots. This invention solves this problem by converting robot position information and safe radii into pseudo-radar data.

[0089] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. An environmental perception method for autonomous navigation of distributed heterogeneous multi-robot systems, characterized in that, Includes the following steps: S1: Initialize robot H, acquire its own 2D radar information, position information, and real-time safe radius, and then broadcast its own position information and real-time safe radius; robot H's position information includes coordinate information in the global coordinate system. Information on the angle between the robot H's forward direction and the positive direction of the global coordinate system. ; S2: Calculate the pseudo radar data of other robots relative to robot H using the position information of robot H, the position information of other robots, and the safety radius of other robots; S3: Integrate the pseudo radar data of other robots relative to robot H and the radar information of robot H itself to obtain the state representation of robot H's environmental perception, and use the state representation of robot H as the state space of robot H. S4: Input the state space into the neural network of the control strategy, train the neural network model through reinforcement learning, and output robot H action a. Robot H executes action a. S5: Determine whether robot H has reached the target point. If it has reached the target point, navigation ends. If it has not reached the target point, continue to step S1. Step S2, which calculates the pseudo radar data of other robots relative to robot H using robot H's position information, other robot position information, and other robots' safety radii, includes the following steps: S21: Calculate the distance and angle between robot H and other robots based on their position information and the position information of the other robots. The position information of the other robots is as follows: The safety radius is The calculation method is as follows: ， in It is the angle between the line connecting robot H and other robots and the positive direction of the global coordinate system. It is the distance between robot H and other robots; S22: Establish a Cartesian coordinate system with robot H as the center and the forward direction of robot H as the positive x-axis. The angle between the line connecting other robots in the robot H coordinate system and the positive x-axis is: ， The equation of the circle formed by the other robot safety radii in the robot's H coordinate system is: ， Transform the equation of the curve into polar coordinates, and let: Therefore, the curve equations for other robots in the polar scale system are: ， Solving the equation yields: S23: The arc formed by the safety radius of other robots is used as the outline of other robots to indicate the perception information of robot H. The equation of the arc curve formed by the safety radius of other robots is: ， The range of values ​​for the equation of the circular arc curve is: The calculation formula is: ， ， Starting from robot H, draw a tangent line that forms a circle with the safety radii of the other robots. Let H be the angle formed by the line connecting robot H and the other robots and this tangent; S24: Assuming robot H's radar scan detects the arc formed by the safety radii of other robots, based on the azimuth angle of robot H's 2D radar information... Find the pseudo-radar data for the azimuth angle of robot H, if the distance between other robots and robot H is less than... ,but ， Among them, the distance between the other robots and robot H is less than hour, Represents the azimuth angle of the i-th robot relative to robot H. Pseudo-radar data at the location, The maximum radar detection radius of robot H is represented by the distance between the i-th robot and robot H, which is greater than the maximum radar detection radius of robot H. Then the pseudo radar data for the azimuth angle corresponding to robot H are all Therefore, the pseudo-radar data for the azimuth angle of the i-th robot relative to robot H is finally: ， in Represents the azimuth angle of the i-th robot relative to robot H. The pseudo-radar data at point i, therefore the pseudo-radar data of the i-th robot relative to robot H can be represented as: ; S25: Fusion of pseudo-radar data of all other robots relative to robot H, i.e., finding the minimum value of the upper radar data of other robots relative to robot H at the corresponding azimuth angle: ， in This represents the azimuth angle of the i-th robot at robot H. Pseudo-radar data at the location, Represents the azimuth angle of all other robots relative to robot H. The minimum value of the pseudo-radar data at that location, therefore the fused pseudo-radar data of other robots relative to robot H is: .

2. The environmental perception method for autonomous navigation of distributed heterogeneous multi-robot systems according to claim 1, characterized in that, In step S1, the 2D radar information of robot H includes distance. and azimuth The robot H position information includes coordinate information in the global coordinate system. Information on the angle between the robot H's forward direction and the positive direction of the global coordinate system. .

3. The environmental perception method for autonomous navigation of distributed heterogeneous multi-robot systems according to claim 2, characterized in that, In step S3, pseudo radar data from other robots relative to robot H and radar information from robot H itself are fused: L fusion This represents the state representation of the robot's environmental perception after fusing pseudo-radar data from H with its own 2D radar distance information.

Images